Technical supplement Severe mental illness and physical ... · The 10 physical health conditions...

Technical supplement Severe mental illness and physical health inequalities

Technical supplement: Severe mental illness and physical health inequalities

2

About Public Health England

Public Health England exists to protect and improve the nation’s health and wellbeing,

and reduce health inequalities. We do this through world-leading science, knowledge

and intelligence, advocacy, partnerships and the delivery of specialist public health

services. We are an executive agency of the Department of Health and Social Care,

and a distinct delivery organisation with operational autonomy. We provide

government, local government, the NHS, Parliament, industry and the public with

evidence-based professional, scientific and delivery expertise and support.

Public Health England

Wellington House

133-155 Waterloo Road

London SE1 8UG

Tel: 020 7654 8000

www.gov.uk/phe

Twitter: @PHE_uk

Facebook: www.facebook.com/PublicHealthEngland

Prepared by: Kate Lachowycz, Sulia Celebi, Gabriele Price, Cam Lugton and Rachel

Roche

For queries relating to this document, please contact: [email protected]

© Crown copyright 2018

You may re-use this information (excluding logos) free of charge in any format or

medium, under the terms of the Open Government Licence v3.0. To view this licence,

visit OGL. Where we have identified any third party copyright information you will need

to obtain permission from the copyright holders concerned.

Published September 2018

PHE publications PHE supports the UN

gateway number: 2018395 Sustainable Development Goals

http://www.gov.uk/phe

https://twitter.com/PHE_uk

http://www.facebook.com/PublicHealthEngland

mailto:[email protected]

https://www.nationalarchives.gov.uk/doc/open-government-licence/version/3/

https://www.gov.uk/government/collections/sustainability-and-public-health-a-guide-to-good-practice


3

Contents

About Public Health England 2

Summary 4

Introduction 5

Methods 6

Statistical analysis 11

Strengths and limitations 12

Appendix 1 15

Appendix 2 16

Appendix 3 18

References 26


4

Summary

This supplement provides technical information on the methods used to examine

inequalities in physical health between people with a diagnosis of severe mental illness

(SMI) and all patients in GP care. It should be used together with Public Health

England’s (PHE’s) research report Severe mental illness (SMI) and physical health

inequalities which provides details on:

background to this work

key results from the analysis

discussion of the findings

summary of interventions and guidance to improve physical health that the

findings support

conclusions

This analysis uses The Health Improvement Network (THIN) database. THIN is a large

general practice database covering about 6% of the UK population. The records in

THIN are anonymised, and longitudinal data is available. Using quality and outcomes

framework (QOF) rules allow a comparison between the prevalence reported by GP

practices and the prevalence derived from THIN. Therefore, the purpose of this

technical supplement is to provide:

detail on how SMI patients are identified in THIN

detail on how all patients (as a comparison group) are identified in THIN

detail on how the examined physical health conditions (including asthma, atrial

fibrillation, cancer, coronary heart disease, chronic obstructive pulmonary

disease, diabetes, heart failure, hypertension, obesity and stroke) are identified

in THIN

a full list of Read of codes used in the analysis

detail on the statistical analysis applied

comprehensive tables of results from the statistical analysis

strengths and limitation of THIN, and the methodology used in the analysis

a basis for a similar sub-national analysis for local areas where access to

general practice data is available

https://www.gov.uk/government/publications/severe-mental-illness-smi-physical-health-inequalities/severe-mental-illness-and-physical-health-inequalities-briefing



5

Introduction

People with SMI, such as schizophrenia and bipolar disorder, are at a greater risk of

poor physical health and have a higher premature mortality than the general

population1. People with SMI in England:

die on average 15 to 20 years earlier than the general population2

have 3.7 times a higher death rate for ages under 752

experience a widening gap in death rates over time3

It is estimated that 2 in 3 deaths in people with SMI are from physical illnesses that can

be prevented4. Major causes of death in people with SMI include chronic physical

medical conditions such as cardiovascular, respiratory, diabetes and hypertension5. In

addition to chronic physical medical conditions, suicide is also an important cause of

death in the SMI population. Suicide risk in people with SMI is high following acute

psychotic episodes and psychiatric hospitalisation. It peaks during psychiatric hospital

admission and shortly after discharge6. Other causes of death include substance abuse,

Parkinson's disease, accidents, dementia (including Alzheimer's disease) and infections

and respiratory acute conditions (particularly pneumonia)5.

In England, there have been few large-scale studies in primary care that look at the

nature and extent of physical health and health inequality in people with SMI7:

experiencing both a mental illness and a physical illness at the same time (known

as co-morbidity)

experiencing a mental illness and more than one physical condition at once

(known as multi-morbidity)

Previous studies in England focused on specific conditions, geographical regions or

looked at overall patterns of inequality (with deprivation largely only considered).7

General practice data is a rich source of longitudinal data on patients. There are a

number of sources of general practice data and research databases, including:

Clinical Practice Research Datalink (CPRD)8

The Health Improvement Network (THIN) database9

bespoke extracts of data using the General Practice Extraction Service

(GPES)10

Public Health England (PHE) currently (May 2018) holds the THIN database. Therefore,

use of THIN is preferable to obtaining other similar datasets that are available for such

research. THIN offers an opportunity to investigate which physical health conditions

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5579328/


6

have a high prevalence among those with SMI known to primary care in England and

how this compares with the general population. Data access to THIN allows analysis at

England-level only. Consequently, the PHE’s Severe mental illness (SMI) and physical

health inequalities report does not provide sub-national findings. However, where local

agreements for access to primary-care data exist the methodology and list of codes

below will allow similar analysis to be carried out.

Findings from the analysis, together with discussion and conclusions, are available in

PHE’s Severe mental illness (SMI) and physical health inequalities report.

Methods

Using THIN data (May 2018, Version: 1705) for GP practices in England, the aim of the

analysis is to examine:

the recorded prevalence of SMI in England by age, sex and deprivation

the proportion of people with recorded SMI in England experiencing co-

morbidities and multi-morbidities

inequalities in the co-morbidities and multi-morbidities between SMI and all

patients in England by age, sex and deprivation

Analysis by age group and sex in people with SMI can help to identify further

inequalities in physical health within patients with SMI. It is also important to consider

the role of socio-economic factors. People experiencing mental health problems

together with long-term physical conditions disproportionately live in deprived areas.11

Deprivation is therefore an important factor in understanding differences in the

prevalence of physical health conditions among those with and without SMI. It is

potentially both a confounder (an explanation for the observed differences between SMI

and all patients) and a moderator (it affects the strength of the relationship for example

between SMI and physical health conditions).

This analysis was approved by the THIN scientific review committee (4 June 2018, SRC

Reference Number 18THIN021).

Description of THIN

THIN9 is a UK primary care database of anonymised electronic medical records

collected in a sample of general medical practices throughout the UK. Extracts are

taken from GP systems every four months. THIN data collection began in 2002 and by

July 2017 over 780 practices had contributed data12. Across the UK there are around

375 GP practices active in THIN and just under 3 million active patients (around 6% of

the general population).





7

THIN data includes non-identified data about:

demographics, such as gender, year of birth and registration dates

consultations, diagnoses and symptoms entered by GPs

prescribed medication

lab tests and results ordered by GP

vaccinations

additional health data such as blood pressure, height and weight

The THIN sample is broadly representative of the UK general population.13 It has similar

demographics, although the THIN sample contains fewer people aged under 25 and is

slightly less deprived. The prevalence of major conditions is comparable to that

recorded by the Quality and Outcomes Framework (QOF) and death rates are close to

national death rates after adjusting for demographics and deprivation13. The GP

recorded SMI prevalence in THIN is broadly comparable with that in other studies and

shows similar patterns by socio-demographic characteristics.14

Patient inclusion criteria

The analysis uses data extracted from THIN in May 2018. The analysis looks at

permanently registered active patients aged 15 to 74 across 154 GP practices in

England. Active patients are patients who are alive and contribute data to THIN.

Patients are only included if they have a complete registration date recorded in THIN

and:

patient flag is either ‘A’ or ‘C’ indicating that the patient record passed the internal

THIN validation process and is an ‘acceptable record’ or is an ‘acceptable record

and transferred out of data due to patient being dead’

registration flag is either ‘01’, ‘02’, ‘05’ or ‘99’ indicating that a patient has

‘Applied’ or is ‘Permanent’ residence or patient has ‘transferred out’ of the

practice or patient has died’

The criteria and definitions to identify patients for the analysis:

only active patients in the age group 15 to 74 are included. Few people aged less

than 15 years have a diagnosis of SMI and focussing on those aged less than 75

years is consistent with the measures of excess mortality in this group5

excluding older age groups also helps avoid mortality attrition bias (earlier deaths

for people with SMI who have physical health conditions)

the analysis looks at cross-sectional point prevalence on the census date (May

2018), with ‘current’ prevalence defined as an unresolved diagnosis. Any historic

diagnosis is included as long as it is not subsequently coded as resolved


8

there is no minimum period of record history for inclusion in the analysis and all

active permanent patients in THIN are included

‘SMI patients’ are defined as having schizophrenia, affective disorder (bipolar or

unspecified affective disorder) or other types of psychoses using Quality and

Outcomes Framework (QOF) business rules15 using Read codes. Patients who

have had a lithium prescription in the preceding six months are not included

although they are included in QOF. Lithium is a medication used to treat SMI and

most commonly bipolar disorder. Full list of Read codes used in the analysis are

available in Appendix 1

in line with QOF business rules for prevalence, counts with a SMI diagnosis ‘In

remission’ are included. ‘In remission’ coding is used by GPs for patients where

these is no record of anti-psychotic medication, mental health in-patient episodes

or secondary care mental health follow-up for at least five years. These patients

remain on the QOF register but are excluded from the quality indicators

similarly to ‘In remissions’ codes, ‘Exception codes’ can be used by GPs to

exclude patients on a QOF register from quality indicators where appropriate.

‘Exception codes’ are not considered in this analysis

‘all patients’ are defined as all active patients aged 15 to 74 and include the SMI

population within it

the Townsend quintiles are allocated to patients based on the Townsend score of

the small geographic area of their home postcode12

deprivation quintiles are available for 78.5% of SMI patients and 75.3% for all

THIN patients. The remaining patients have no allocated score as their postcode

cannot be linked to Townsend score or a score is not available due to data

quality reasons

The 10 physical health conditions for SMI and all patients in THIN identified using QOF

business rules15 are:

asthma – patients with a history of asthmaa

atrial fibrillation (AF) – patients with an atrial fibrillation event, including

paroxysmal, persistent and permanent

cancer – patients with a diagnosis of cancer excluding non-melanoma skin

cancers

coronary heart disease (CHD) – patients who have had coronary artery

revascularisation procedures, such as coronary artery bypass grafting (CABG)

chronic obstructive pulmonary disease (COPD) – patients with symptoms of a

persistent cough, sputum production, or dyspnoea and/or a history of exposure to

a QOF excludes patients without any prescription of asthma medication in the previous 12

months

https://qof.digital.nhs.uk/

https://qof.digital.nhs.uk/


9

risk factors for the disease, and diagnosis confirmed by post bronchodilator

spirometry

diabetes – all patients with diabetes mellitus (type 1 and type 2)

heart failure (HF) – all patients with a diagnosis of HF

hypertension – patients with established hypertension, for example, elevated

blood pressure readings of greater than 140/90 mmHg on three separate

occasions

obesity – patients with a BMI ≥30 in the preceding 12 months

stroke – patients with a diagnosis of stroke or transient ischaemic attack (TIA)

As the 10 conditions included are a subject of QOF registers:

diagnosis recording is likely to be good

the results can be checked against the QOF figures

the analysis can easily be reproduced

The analysis uses full patient history to define co-morbidities and multi-morbidities.

Given the low prevalence of SMI in primary care, this maximises the sample size

available for the analysis. Consistent with QOF, patients have a specific physical

condition if they have a Read-code diagnosis that is not subsequently coded as

‘resolved’. Full list of Read codes used in the analysis is available in Appendix 1.

Details of variables included in the data extraction and a summary of inclusion criteria

are available in Appendix 2.

Validation of prevalence

To assess the accuracy of SMI and physical conditions prevalence from THIN, this

analysis compares the prevalence recorded in THIN with the prevalence recorded in

QOF for the period April 2016 to March 201716. This is currently the latest available data

for QOF.

The SMI prevalence in THIN is lower than recorded in QOF (Table 1). This is partially

due to lithium prescribing being not included in THIN analysis. For the majority of the

ten examined physical health conditions, the prevalence compares well with the

England QOF prevalence. The prevalence figures in THIN are within 0.5 percentage

points of the QOF value with greater differences recorded for asthma, hypertension and

obesity (Table 1).


10

Table 1 Condition prevalence (%) in THIN (May 2018) and QOF 2016 to 2017

Note the percentages are not age and sex standardised. * age groups used to match prevalence in each specific QOF register

Source: The Health Improvement Network (THIN). Data extracted May 2018; NHS Digital; Quality and Outcomes Framework – Prevalence, Achievements and Exceptions Report. England, 2016-17

17

For asthma, QOF excludes patients without any prescription for asthma medication in

the previous 12 months. This analysis captures all people with an unresolved diagnosis

of asthma and possibly resulting in a higher prevalence of asthma. Obesity and

hypertension shows a large variation in QOF prevalence between sub-regions across

England18. Therefore, the location of GP practices in THIN sample can affect the

recorded prevalence for both conditions. Demographic and deprivation profiles between

the QOF and the THIN population may also play a role. This analysis does not account

for the differences.

Condition (ages* included)

THIN prevalence % (95% CI)

England QOF Prevalence %

Asthma (all ages) 9.00 (8.95-9.05) 5.94

Atrial Fibrillation (all ages) 1.64 (1.62-1.66) 1.84

Cancer (all ages) 2.99 (2.96-3.02) 2.58

CHD (all ages) 2.70 (2.68-2.73) 3.15

COPD (all ages) 1.59 (1.57-1.61) 1.87

Diabetes (over 17 years) 6.22 (6.18-6.27) 6.67

Heart Failure (all ages) 0.86 (0.85-0.88) 0.79

Hypertension (all ages) 12.40 (12.34-12.45) 13.83

Obesity (over 18 years) 7.40 (7.35-7.45) 9.65

SMI (all ages) 0.73 (0.71-0.74) 0.92

Stroke & TIA (all ages) 1.47 (1.45-1.49) 1.75


11

Statistical analysis

To account for demographic differences between SMI patients and all patients, this

analysis applies standardisation for age and sex using the direct method19 to the

prevalence of physical health conditions in SMI patients. This analysis applies the age

and sex specific prevalence observed in the SMI patients to the age structure of the

general THIN population. In other words, the age and sex specific prevalence observed

in SMI patients were applied to the age structure of all patients in THIN. This analysis

also uses direct standardisation to adjust for deprivation. To allow for easy replicability

of the analysis, 95% confidence intervals (CIs) are used to determine statistical

significance. However, this approach is more conservative than formal testing and the

use of p-values20. The use of CIs to access statistical significance rather than p-value is

widely accepted and currently used across PHE’s intelligence products. Both are not

contradictory statistical concepts and can be complementary. The 95% CIs calculated in

this analysis use Wilson Score and Dobson’s methods as recommend by PHE21 for

proportions and directly standardised rates respectively. Differences in the prevalence

are considered as statistically significant if the 95% CIs do not overlap. All patient

prevalence is used as a reference (point) value and therefore does not include CIs. All

comparisons were at an England level. The rate ratio of prevalence is calculated by

dividing SMI patient prevalence by all patient prevalence. A rate ratio higher than 1

indicates higher prevalence in SMI patients.

This analysis compares SMI patients to all patients aged 15 to 74 in THIN to examine if:

the prevalence of each of the 10 physical health conditions differs between both

patient groups

the prevalence of co-morbidities and multi-morbidities differs between both

patient groups

there are further differences by sex and age group in 20-year bands (15 to 34,

35 to 54, 55 to 74), and socio-economic deprivation

Analysis by sub-group is only carried out where there is sufficient sample size to allow a

robust comparison.


12

Strengths and limitations

This analysis has a number of strengths and limitations that relate to the THIN database

itself and this methodological approach. Those should be kept in mind when interpreting

the results from the analysis and repeating the analysis at local level. THIN data

THIN data has a number of strengths around its use in population level and

epidemiological studies [^22] such as:

it is a very large data set containing records of approximately 6% of UK

population

it is broadly representative of the UK population

it allows patients with particular conditions to be identified and an analysis in

relation to the conditions to be carried out

it can be used to select control subjects from the same source population

data is collected in a non-interventional way and therefore reflects ‘real life’

information is continually updated, permitting investigation of the effects of new

interventions and treatments

data does not have to be collected which may reduce time and costs

it can be used in most epidemiological study designs (that is cohort, case-control,

case-series).

it can be used to study relatively rare exposures or outcomes

a number of practices have been linked to hospital episode statistics (HES) data

allowing the analysis of both primary and secondary care data

There are some limitations around THIN data because [^22]:

the primary use of Vision software that is used for THIN is patient management

and not medical research, and data will reflect only those events that are deemed

to be relevant to the patient’s care

despite being a large data set THIN may still have power problems when

exposure and outcomes are both rare

as THIN is a sample of the GP practice population, when no-one is identified as

having a specific condition, it cannot be assumed that the count is actually zero

it may be inappropriate for use in studies where individual ethnicity, occupation,

employment, and/or socio-economic status are important variables (not available

at patient-level data)

it may be inappropriate for use in studies on drug-related exposures as non-

compliance to medication prescriptions may be an issue


13

it may be inappropriate for use in studies where data are primarily related to

secondary care (cancer-care studies)

it may be inappropriate for use in studies examining over-the-counter drugs as

these will not usually be recorded

it may be inappropriate for use in studies looking at lab test results before

computerisation as only abnormal values may have been entered

the number of GP practices subscribing to THIN is showing a downward trend

.

Few of the above limitations apply to this analysis.

Methods

The strengths of the method for this analysis are:

QOF business rules are used and therefore coding is likely to be good

use of QOF allows validation of findings against other published data

confirmed diagnoses were used for physical health conditions therefore reducing

the effect of misdiagnosis

use of ‘in remission’ code as this allows inclusion of patients with SMI who are

not included in QOF quality indicators but are still likely to experience poor

physical health

use of ages 15 to 74 to account for the effect of premature morality in patients

with SMI especially in relation to standardisation of the data

inclusion of a range of physical health conditions

examination of overall inequalities between SMI and all patients together with the

effects of age, sex and deprivation

The methodology used in this analysis has a number of limitations as:

it is based on diagnosed conditions only and therefore cannot account for levels

of undiagnosed physical health conditions in SMI or all patients that could

potentially increase the actual health inequality reported in this analysis

it relies entirely on accurate diagnostic coding although coding across GP

practices contributing to THIN is likely to be good

GP practices contributing to the THIN database might not be fully representative

of all GP practices in England in their coding behaviour

it does not include antipsychotic medication prescribing in addition to SMI

diagnosis to determine prevalence

only individuals who are registered with GPs can be included in the analysis and

mental health problems are likely to be an issue in parts of the population with no

GP registration such as homeless individuals


14

there are some difference in QOF and Read codes used in this analysis

looking at diagnosis of a condition does not tell us anything about subsequent

treatments and interventions that the patient received

it does not look at differences in physical health conditions by type of SMI

diagnosis, for example schizophrenia and affective disorder

the analysis is cross-sectional, and it is not possible to look at the temporal

association (the order in which conditions were diagnosed) of SMI and physical

health conditions or consider if there is a ‘healthy survivor’ bias, where people

with the poorest health die early

using currently active patients results in a small sample of people with both SMI

and less common physical health conditions such as cancer, atrial fibrillation,

stroke and heart failure


15

Appendix 1

Table 2 Read codes for types of physical health conditions and SMI

Condition Condition group Diagnosis codes Resolved

codes

Asthma Respiratory H33.% (excluding H333), H3120, H3B, 173A 21262, 212G

Atrial fibrillation

Cardiovascular G573.% 212R

Cancer Cancer B0-B32z, B34-B6z0 (excluding B677), Byu-Byu41, Byu5-ByuE0, K1323, K01w1, 68W24, C184

Coronary heart disease (CHD)

Cardiovascular G3-G309, G30B-G330z (excluding G310), G33z-G3401, G342-G35X, G38-G3z, Gyu3.% (excluding Gyu31)

Chronic obstructive pulmonary disease (COPD)

Respiratory H3, H31.% (excluding H3101, H31y0, H3122), H32.%, H36-H3z (excluding H3y0, H3y1), H5832, H4640, H4641, Hyu30, Hyu31

2126F

Diabetes Long-term conditions

C10, C109J, C109K, C10C, C10D, C10E.%, C10F % (excluding C10F8), C10G.%, C10H.%, C10M.%, C10N.%, PKyP, C10P.%, C10Q

21263, 212H

Heart failure Cardiovascular G58.%, G1yz1, 662f.-662i

Hypertension Cardiovascular G2, G20.%, G24-G2z (excluding G24z1, G2400, G2410, G27), Gyu2, Gyu20 21261, 212K

Obesity Lifestyle 22K5, 22K7, 22KC, 22KD, 22KE 22K (with BMI 30+)

Serious mental illness (SMI)

Mental health E10.%, E110.%, E111.%, E1124, E1134, E114-E117z, E11y.% (excluding E11y2), E11z, E11z0, E11zz, E12.%, E13.% (excluding E135), E212, Eu2.%, Eu30.%, Eu31.%, Eu323, Eu328, Eu333, Eu32A, Eu329

Stroke Cardiovascular G61.% (excluding G617), G63y0-G63y1, G64.%, G66.% (excluding G669), G6760, G6W., G6X., Gyu62-Gyu66, Gyu6F, Gyu6G, G65-G654, G656-G65zz, ZV12D, Fyu55, G63y0-G63y1, G64.%, G665, G666, G6760, G6W, G6X, Gyu63-Gyu66, Gyu6G


16

Appendix 2

Table 3 Variables and inclusion criteria for THIN data extraction

Variables Criteria

Selecting active

patients

Unique combined patient identifier

Sex

Age

Unique patient identifier within GP

practice

Townsend score (most recently

recorded score)

Valid demography record

Permanent patient registration

Active practice in England

Aged 15-74

Co-morbidities

master – a complete

diagnosis (all

conditions)

Unique combined patient identifier

Sex

Age


recorded score)

Read code

Event date

Condition

Condition category

Read code type (diagnosis/resolved)

QOF age for condition

BMI

Condition number (distinct conditions

numbered by event date with the

most recent event recorded as 1)

Medical record integrity is acceptable

Non-missing event date

Condition presence based on eligible

read code and age restrictions as

defined by QOF Business Rules

Co-morbidities Unique combined patient identifier Most recently recorded diagnosis of


17

Variables Criteria

distinct – latest active

diagnosis of each

condition

Sex

Age


recorded score)

Condition

Condition category

QOF age for condition

Diagnosis date (most recent)

Resolved date (if condition was

diagnosed and resolved prior to the

latest diagnosis)

each condition

Active diagnosis (diagnosis event

date must be more recent than

resolved event date)


18

Appendix 3

The below tables provide full results presented in the research report Severe mental

illness (SMI) and physical health inequalities.

Table 4 Age and sex distribution of patients with a diagnosis of SMI compared with all

THIN patients

Key: * = counts below 6

Age group (years)

Number of females (%) Number of males (%)

SMI THIN SMI THIN

0-4 * 41675 (5.80) * 43450 (6.20)

5-9 * 44388 (6.18) * 46931 (6.70)

10-14 * 43555 (6.06) * 44903 (6.41)

15-19 56 (1.09) 38711 (5.39) 54 (1.05) 39793 (5.68)

20-24 206 (4.01) 44611 (6.21) 251 (4.86) 42289 (6.04)

25-29 289 (5.62) 50896 (7.08) 364 (7.05) 48289 (6.89)

30-34 361 (7.02) 51788 (7.21) 473 (9.17) 49431 (7.06)

35-39 402 (7.82) 49555 (6.90) 557 (10.79) 49166 (7.02)

40-44 444 (8.63) 46651 (6.49) 524 (10.16) 48224 (6.88)

45-49 534 (10.39) 50580 (7.04) 639 (12.38) 51957 (7.42)

50-54 558 (10.85) 50704 (7.05) 628 (12.17) 51680 (7.38)

55-59 511 (9.94) 42977 (5.98) 512 (9.92) 44259 (6.32)

60-64 377 (7.33) 35733 (4.97) 351 (6.80) 35701 (5.10)

65-69 400 (7.78) 36119 (5.03) 307 (5.95) 33769 (4.82)

70-74 323 (6.28) 30582 (4.26) 236 (4.57) 27662 (3.95)

75-79 271 (5.27) 22620 (3.15) 136 (2.64) 18975 (2.71)

80-84 193 (3.75) 18128 (2.52) 77 (1.49) 13614 (1.94)

85-89 137 (2.66) 11931 (1.66) 32 (0.62) 7339 (1.05)

90+ 76 (1.48) 7496 (1.04) 17 (0.33) 3163 (0.45)

Total 5142 718700 5160 700595




19

Table 5 Prevalence of SMI by sex, age group and deprivation patients aged 15 to 74

Table 5 Prevalence (age and sex standardised) of physical health conditions for SMI

and all patients aged 15 to 74

* Rate ratio = prevalence in SMI/prevalence in all patients. Bold indicates if SMI

prevalence is significantly higher

Sub-group Number of people with SMI

Number of people in THIN

Prevalence as % (95% Confidence Intervals - CIs)

Se

x Female 4461 528907 0.84 (0.82-0.87)

Male 4896 522220 0.94 (0.91-0.96)

Ag

e g

rou

p

(ye

ars

)

15-34 2054 365808 0.56 (0.54-0.59)

35-54 4286 398517 1.08 (1.04-1.11)

55-74 3017 286802 1.05 (1.02-1.09)

Dep

riv

ati

on

qu

inti

le

Least deprived 1204 221301 0.54 (0.51-0.58)

1060 160233 0.66 (0.62-0.70)

1548 169934 0.91 (0.87-0.96)

1821 144240 1.26 (1.21-1.32)

Most deprived 1709 95678 1.79 (1.70-1.87)

Condition Prevalence (%) in SMI patients (95% CIs) Prevalence (%) in THIN

patients Rate ratio*

Obesity 12.82 (12.12-13.55) 7.03 1.8

Asthma 12.35 (11.48-13.26) 10.02 1.2

Hypertension 12.05 (11.41-12.72) 11.50 1.0

Diabetes 9.51 (8.92-10.12) 4.94 1.9

COPD 2.90 (2.59-3.24) 1.39 2.1

Cancer 2.42 (2.13-2.73) 2.66 0.9

CHD 2.42 (2.13-2.73) 2.02 1.2

Stroke 1.59 (1.36-1.85) 0.98 1.6

Atrial fibrillation 0.86 (0.70-1.06) 0.95 0.9

Heart failure 0.82 (0.65-1.01) 0.55 1.5


20

Table 6 Age-specific prevalence of physical health conditions by age group for SMI

patients and all patients aged 15 to 74

* Rate Ratio = prevalence in SMI/prevalence in all patients. Bold indicates if SMI

prevalence is significantly higher.

Condition Age group

(years) Prevalence (%) in SMI

patients (95% CIs) Prevalence (%) in THIN patients

Rate Ratio*

Obesity

15-34 7.01 (5.98-8.20) 2.33 3.0

35-54 15.10 (14.06-16.20) 8.11 1.9

55-74 18.66 (17.31-20.09) 11.51 1.6

Asthma

15-34 15.00 (13.52-16.60) 11.85 1.3

35-54 11.99 (11.05-13.00) 9.23 1.3

55-74 8.95 (7.98-10.02) 8.77 1.0

Hypertension

15-34 1.36 (0.94-1.96) 0.43 3.2

35-54 9.75 (8.90-10.68) 7.66 1.3

55-74 29.40 (27.80-31.05) 30.95 0.9

Diabetes

15-34 2.39 (1.81-3.14) 0.65 3.7

35-54 9.80 (8.94-10.73) 3.77 2.6

55-74 18.93 (17.57-20.36) 12.04 1.6


21

Table 7 Prevalence (age standardised) of physical health conditions by sex for SMI and

all patients aged 15 to 74

Sex Condition Prevalence (%) in SMI

patients (95% CIs) Prevalence (%) in

THIN patients Rate

Ratio*

Fem

ale

Obesity 13.94 (12.91-15.03) 7.89 1.8

Asthma 13.87 (12.53-15.30) 10.36 1.3

Hypertension 11.34 (10.48-12.26) 10.71 1.1

Diabetes 8.69 (7.89-9.56) 4.03 2.2

COPD 2.62 (2.22-3.08) 1.31 2.0

Cancer 2.94 (2.51-3.43) 3.16 0.9

CHD 1.36 (1.07-1.70) 1.08 1.3

Stroke 1.44 (1.14-1.79) 0.81 1.8


Heart failure 0.67 (0.48-0.92) 0.39 1.7

Ma

le

Obesity 11.68 (10.75-12.67) 6.15 1.9

Asthma 10.81 (9.71-11.97) 9.66 1.1

Hypertension 12.77 (11.83-13.77) 12.30 1.0

Diabetes 10.34 (9.49-11.23) 5.86 1.8

COPD 3.19 (2.72-3.71) 1.47 2.2

Cancer 1.89 (1.53-2.31) 2.17 0.9

CHD 3.49 (3.00-4.05) 2.97 1.2

Stroke 1.74 (1.40-2.15) 1.15 1.5


Heart failure 0.96 (0.71-1.27) 0.72 1.3




22

Table 8 Prevalence (age, sex and deprivation standardised) of physical health

conditions for SMI and all patients aged 15 to 74



Condition Prevalence (%) in SMI

patients (95% CIs) Prevalence (%) in THIN

patients Rate

Ratio*

Obesity 12.86 (11.98-13.78) 7.03 1.8

Asthma 12.01 (11.11-12.95) 10.02 1.2

Hypertension 12.02 (11.22-12.85) 11.50 1.0

Diabetes 9.36 (8.65-10.10) 4.94 1.9

COPD 2.58 (2.25-2.95) 1.39 1.9

Cancer 2.85 (2.45-3.28) 2.66 1.1

CHD 2.65 (2.26-3.08) 2.02 1.3

Stroke 1.74 (1.43-2.08) 0.98 1.8


23

Table 9 Prevalence (age and sex standardised) of physical health conditions (obesity,

asthma, hypertension and diabetes) by deprivation quintile for SMI and all patients aged

15 to 74



Condition Deprivation

quintile Prevalence (%) in SMI

patients (95% CIs) Prevalence (%) in

THIN patients Rate

Ratio*

Obesity

Least deprived 13.80 (11.78-16.06) 6.43 2.1

12.29 (10.28- 14.59) 6.72 1.8

12.90 (11.15-14.82) 7.02 1.8

11.80 (10.29-13.47) 6.73 1.8

Most deprived 13.15 (11.14-15.05) 7.49 1.8

Asthma

Least deprived 9.62 (7.86-11.64) 10.57 0.9

11.04 (8.97-13.44) 10.19 1.1

12.89 (11.01-14.98) 10.43 1.2

14.31 (12.39-16.41) 10.60 1.4

Most deprived 14.15 (12.10-16.43) 11.12 1.3

Hypertension

Least deprived 11.64 (9.85-13.66) 12.90 0.9

12.37 (10.47-14.51) 12.70 1.0

13.10 (11.48-14.88) 12.07 1.1

11.03 (9.70-12.49) 11.08 1.0

Most deprived 11.88 (10.44-13.46) 11.50 1.0

Diabetes

Least deprived 7.96 (6.46-9.70) 4.47 1.8

9.86 (8.12-11.86) 4.84 2.0

9.39 (8.03-10.93) 5.33 1.8

9.01 (7.81-10.35) 5.74 1.6

Most deprived 12.19 (10.62-13.92) 6.62 1.8


24

Table 10 Prevalence (age and sex standardised) of physical health conditions (COPD,

cancer, CHD and stroke) by deprivation quintile for SMI and all patients aged 15 to 74

*

Rate Ratio = prevalence in SMI/prevalence in all patients. Bold indicates if SMI


Condition

Deprivation quintile

Prevalence (%) in SMI patients (95% CIs)

Prevalence (%) in THIN patients

Rate Ratio*

COPD

Least deprived 1.28 (0.73-2.06) 0.94 1.4

2.57 (1.77- 3.61) 1.21 2.1

3.26 (2.49-4.20) 1.57 2.1

3.45 (2.75-4.28) 1.89 1.8

Most deprived 3.12 (2.44-3.94) 2.44 1.3

Cancer

Least deprived 3.66 (2.71-4.84) 3.53 1.0

3.44 (2.49-4.62) 3.17 1.1

2.34 (1.69-3.16) 2.67 0.9

2.16 (1.59-2.85) 2.29 0.9

Most deprived 1.90 (1.32-2.64) 2.01 0.9

CHD

Least deprived 3.39 (2.39-4.66) 2.06 1.6

2.39 (1.61-3.42) 2.18 1.1

2.84 (2.13-3.71) 2.19 1.3

2.02 (1.49-2.68) 2.19 0.9

Most deprived 1.95 (1.43-2.61) 2.45 0.8

Stroke

Least deprived 2.02 (1.29-3.00) 1.00 2.0

1.49 (0.89-2.32) 1.01 1.5

2.22 (1.59-3.01) 1.09 2.0

1.24 (0.83-1.77) 1.10 1.1

Most deprived 1.40 (0.92-2.04) 1.18 1.2


25

Table 11 Prevalence of physical health co-morbidities for SMI and patients aged 15 to

74



Table 12 Prevalence (age and sex standardised) of 3 and more physical health co-

morbidities for SMI and all patients aged 15 to 74



Number of conditions Prevalence (%) in SMI

patients (95% CIs) Prevalence (%) in THIN

patients Rate Ratio*

Any physical health condition

41.44 (40.45-42.45) 29.50 1.4

2 or more 15.57 (14.85-16.32) 8.73 1.8

3 or more 5.07 (4.64-5.53) 2.74 1.8

4 or more 1.62 (1.39-1.90) 0.80 2.0

5 or more 0.41 (0.30-0.56) 0.2 1.9

Sub-group

Prevalence (%) in SMI patients (95% CIs)

Prevalence (%) in THIN patients

Rate Ratio*

Se

x Female 4.03 (3.52-4.60) 2.40 1.7

Male 4.67 (4.10-5.30) 3.08 1.5

Ag

e g

rou

p

(ye

ars

)

15-34 0.44 (0.23-0.83) 0.09 5.1

35-54 2.99 (2.52-3.54) 1.26 2.4

55-74 11.17 (10.10-12.34) 8.18 1.4

Dep

riv

ati

on

qu

inti

le

Least deprived 3.83 (2.82-5.07) 2.54 1.5

3.27 (2.33-4.45) 2.68 1.2

4.97 (3.99-6.11) 2.90 1.7

4.08 (3.32-4.97) 3.15 1.3

Most deprived 4.30 (3.48-5.26) 3.68 1.2


26

References 1 DE Hert M and others. ‘Physical illness in patients with severe mental disorders’ World Psychiatry 2011: volume 10, issue 1, pages 52-77 (viewed on 2 July 2018) 2 Chesney E and others. ‘Risks of all-cause and suicide mortality in mental disorders: a meta-review’ World Psychiatry 2014: volume 13, issue 2, pages 153-160 (viewed on 2 July 2018) 3 Hayes JF and others. ‘Mortality gap for people with bipolar disorder and schizophrenia: UK-based cohort study 2000-2014’ The British Journal of Psychiatry 2017: volume 211, pages 175-181 (viewed on 2 July 2018) 4 The Mental Health Taskforce, NHS England. ‘Five Year Forward View for Mental Health’ 2016 (viewed on 2 July 2018) 5 John A and others. ‘Premature mortality among people with severe mental illness’ Schizophrenia Research 2018 (available online only) (viewed on 2 July 2018) 6 Appleby L and others. ‘Safety First: Five-year report of the National Confidential Inquiry into

Suicide and Homicide by People with Mental Illness’ 2001 (viewed on 2 July 2018) 7 Reilly, S and others ‘Inequalities in physical comorbidity: a longitudinal comparative cohort study of people with severe mental illness in the UK’ British Medical Journal 2015: vol. 5, issue 12 (available online only) (viewed on 2 July 2018) 8 Clinical Practice Research Datalink. ‘Welcome to Clinical Practice Research Datalink’, accessed 2 July 2018 9 IQVIA. ‘The Heath Improvement Network (THIN)’, accessed 2 July 2018 10 NHS Digital. ‘General Practice Extraction Service’, accessed 2 July 2018 11 Reilly, S, Olier, I, Planner, C et al, 2015. ‘Inequalities in physical comorbidity: a longitudinal comparative cohort study of people with severe mental illness in the UK’. British Medical Journal. vol. 5, no. 12 12 QuintilesIMS. 2017. ‘THIN Data Guide for Researchers. Version: 1705 13 Blak, BT, Thompson, M, Dattani, M, 2011. ‘Generalisability of The Health Improvement Network (THIN) database: demographics, chronic disease prevalence and mortality rates.’ Informatics in Primary Care. vol. 19, issue. 4. p251-255 14 Hardoon, S, Hayes, J F, Blackburn, R,. ‘Recording of Severe Mental Illness in United Kingdom Primary Care, 2000–2010’. PLOS ONE 2013: volume 8, number 12 (available online only) (viewed on 2 July 2018) 15 NHS Digital. ‘Quality and Outcome Framework (QOF) business rules’, accessed 2 July 2018 16 NHS Digital, 2018. ‘Quality and Outcomes Framework (QOF) 2016/17 results’, accessed 2 July 2018 17 NHS Digital, 2017. ‘Quality and Outcomes Framework (QOF) – 2016-17’, accessed 2 July 2018 18 NHS Digital, 2017. ‘Quality and Outcomes Framework – Prevalence, Achievements and Exceptions Report’, accessed 2 July 2018 19 Public Health England, 2017. ‘Technical Guide Calculating Directly Standardised Rates’, accessed 2 July 2018 20

Greenland S and others ‘Statistical tests, P values, confidence intervals, and power: a guide to misinterpretations’ European Journal of Epidemiology 2016: volume 31, pages 337-350 (viewed 2 August 2018) 21 Public Health England. ‘Fingertips: Technical Guidance’, accessed 2 July 2018 22 University College London. ‘THIN data: Strengths & Limitations’, accessed 2 July 2018

http://www.ucl.ac.uk/pcph/research/thin-database/database/pros-cons

Technical supplement Severe mental illness and physical ... · The 10 physical health conditions...

Documents

Transcript of Technical supplement Severe mental illness and physical ... · The 10 physical health conditions...